Supporting self-efficacy beliefs and interest as educational inputs and outcomes: Framing AI and Human partnered task experiences

Abstract

Interest and self-efficacy are crucial to academic success. This study addresses two gaps in our understanding of their development and support during university courses: how prior self-efficacy and interest plays a role in, and how different classroom activities build toward the development of students' future interest and self-efficacy. In this study the interplay between ability-beliefs (self-efficacy/self-concept) and interest at three levels of specificity (Domain, Course and Task) were tested across a Japanese university language course (n = 128). Within this test, students' interest in two language practice tasks (i.e., Human and then Chatbot partners) were assessed and compared. Prior interest was a robust predictor of all future task/course interest. Only Human-Human task interest directly predicted future course self-efficacy, but was mediated by course interest for future domain interest. For future interest, Human practice partners are superior to AIs. Supporting prior domain and later course interest should be a focus for university educators.